Why scientists, vendors and end users are usually wrong about artificial intelligence

Why scientists, vendors and end users are usually wrong about artificial intelligence

Artificial intelligence can be many things - depending on who you’re talking to. To scientists, AI is a work in progress, with decades of research still required to achieve perfection. To software vendors, AI is an existing and quite amazing product, ‘not like anything you’ve seen before’. To end users, it’s usually ‘yet-another-technology’, which may or may not optimize their way of doing business. All of them are right, in a sense and yet, all of them are wrong. And here are the reasons why.

Scientists are right, as artificial intelligence is still years away from what science-fiction fans fancy: a global super-intelligence, able to act like a human being or able to replace a human being without anyone noticing, performing all everyday tasks in a more efficient way than a human. In scientific language, this would be a ‘strong AI’, while today’s AIs are ‘weak AIs’. Unlike strong AIs, they’re designed to fulfill one task, such as classifying and responding to emails, playing a GO-game or driving a vehicle. Scientists are not sure when a strong AI will emerge and they forecast different timelines for that. They also point out that AI's accuracy when performing a task can be improved.

Scientists are right in the sense that a general-purpose AI or even 100% perfect specialized AIs do not exist, but that doesn’t matter, really. As of today, AIs already have the potential to drastically change the way we work and live. Artificial intelligence does not have to be perfect to help humans in their everyday tasks such as managing an agenda, monitoring hundreds of video-surveillance feeds, driving a vehicle, detecting cancer… It might even be better if AI never gets to the state of being ‘strong ‘. Many scientists such as Stephen Hawking are alarmed about the moment when machine’s intelligence would prevail over human’s intelligence. The human race is not prepared for this… yet.

Software vendors are also right in what they are saying, AI is a game changer. The ultimate aim of the entire B2B software industry is to automate procedures and rapidly perform complex computation, through specific software such as ERPs, business rules engines, decision support systems, etc. Despite bold claims, this aim hasn’t been achieved yet. Artificial intelligence, on the other hand, can make this dream come true. Making B2B software work required to capture people’s knowledge and to update it whenever changes would be made. This is a tedious process, even with the best, most user-friendly software. While AI is continuously learning and updating itself, it achieves levels of accuracy never achieved by human beings. For instance, an AI-empowered program can, in a heartbeat, analyze a video, identify the people talking even in a noisy environment, provide distinct transcriptions for each interlocutor, split the video into chapters and provide a summary for each of them. An AI-empowered program can also analyze petabytes of data, schemes, regulations and more, to answer a prospect’s inquiry, with the highest chance possible to transform this inquiry into a profitable lead.

Whether this enthusiasm is justified or not, it creates very high expectations which will inevitably provoke a high level of disappointment. Though AI-empowered software can prove extremely efficient in some circumstances, it cannot resolve every problem like magic. Implementing an AI solution requires a certain level of maturity, regarding the processes to be automated, the way it can help optimize or create services and, globally, the digital transformation strategy. Such a strategy should have a real focus on people and how they’ll be able to leverage a potentially new AI technology, rather than a focus on the technology itself.

End users are right in the sense that, like all new technologies coming to the market, AI solutions provide only slight upgrades to their daily operations. AI-empowered programs may save them a few hours of work here and there, like other technologies have done in the past. An IT support team implementing an AI-empowered self-service portal will get more time to work on complex issues. Doctors or hotel revenue managers will use AI as assistants but will check the results by themselves. In heavy-processed industries, AI will monitor and optimize the processes to raise the level of quality. From this point of view, AI equals optimization, not revolution.

What users don’t see, though, is that AI is changing the whole economy, by accelerating the digital transformation. Impacts may be so important that some jobs or even some businesses may disappear in the short term. The ‘uberization’ of the economy has provoked endless discussions, with taxis protesting against the disruption induced by a digital-empowered user-friendly car riding service. Now, imagine the potential disruption AI-powered autonomous vehicles can bring. Even if some vendors exaggerate about the potential of their AI solution (or even present big data offerings or other software as AI), artificial intelligence will have a deeper impact than many end users imagine.

As usual, when everybody’s right and everybody’s wrong, the truth lies somewhere in between. Scientists can be proud of what they already accomplished, while vendors should be more careful in wielding the term ‘artificial intelligence’. Also, and above all, end users should examine and explore AI capabilities for what they can provide today, in terms of optimization, and how artificial intelligence will impact themselves as an industry, as a business, and as a person. Artificial intelligence will trigger major changes, and everybody should be ready to embrace them – or better, to anticipate them.

Don't miss our latest report on artificial intelligence: 'Demystification and Market Opportunities'